Angels on pinheads

[From Bill Powers (931130.1545 MST)]

Bruce Nevin (931130.1053 EST),
Rick Marken (931129.1300) --

You guys are arguing just to be arguing. The "data" you're
arguing about is a STORY, made up by an author. It is not an
observation about what real people did or felt, but a caricature
created from one person's experience and imagination, with a
heavy emphasis on imagination. Would real people act and react in
these limited, stereotyped, and extreme ways? Why not argue about
what kind of control system Mary Poppins used to fly with an
umbrella?

The trouble with stories is that the author can manipulate the
actors in any way at all. If those were real people, someone
would have been murdered by now.

···

---------------------------------------------------------------
Martin Taylor (931130.1130) --

The purpose you see is still within the organism, not in the
wider system that includes the whole control loop.

The "wider system" includes many entities that are not control
systems. Those entities don't act purposively; they are often
acted upon by living systems to achieve purposes devised by
living systems, but they never act on living systems to achieve a
purpose devised by the environment. Purpose exists in the
universe just as surely (exactly as surely) as electrons exist in
the universe. Maybe the universe as a whole isn't purposive, but
the universe as a whole isn't an electron, either. The universe
as a whole isn't any one thing. So what?

Martin:

I don't mind if you give every vortex in a stream "purpose,"
but I do think that this usage goes a bit beyond the normal
bounds of the term.

Vortices in a stream are not control systems. The energy upon
which they draw to maintain their stability is the same energy
carried by the forces that perturb them. That's not how a control
system works, as you pointed out yourself.

Martin:

Within any framework, any particular explanation may be in
error, and the error found by testing the explanation within
the framework. There is no ground for criticizing an
explanatory framework because the explanations within it do
not explain the things YOU want explained, if they explain the
things someone else wants explained.

This is a statement about mathematical frameworks, not about an
experimental science. In mathematics you state your rules and
then you stick to them, and nobody has any reason to complain
because it's a self-sufficient and closed system. Our main
dispute with the dynamicists rests on the fact that we both
purport to be explaining the same observations, yet we have
arrived at mutually contradictory explanations. So we can't just
let the differences go; that would amount to anarchy, not
science. I claim that the dynamicists are ignoring facts.

Martin:

If a dynamicist has an explanation that works for some
phenomenon he/she is interested in but you are not, that is no
ground for criticizing their framework.

One doesn't have to reject the dynamicists' framework to bring up
counterexamples, to show implications that are contrary to
observation, to point out that disturbances exist in normal
environments and that actions affect inputs at the same time that
inputs are affecting actions. These facts and relationships are
part of the data, not of the theory. Ignoring data is not a
legitimate basis for claiming that a theory is correct, even
within its own framework. Ignoring premises and implications that
are contrary to observation is ... well, it's not how you learn
anything about the world.

Rick:

Their research methodology guarantees that controlled variables
will remain undetected.

Martin:

So what, if their results are interesting to them, and hold
whether or not particular variables are controlled.

Psychologically, you're perfectly right, Martin. People not only
have the right to pursue what is interesting to them and ignore
what is not, but they are so constructed that this is what they
WILL do if that's what they desire to do, no matter how unable
they are to justify their approaches to anyone else. As you say,
if this is how people want to play the game, this is how they
will play it. But they will play it without me.

Martin:

All self-organization depends on negative feedback, and I don't
consider it control unless there is a variable reference level.

I would consider it control if the (negative) loop gain was
greater than 5 or so. All systems in which the perturbing event
supplies the energy that is then used to restore equilibrium have
loop gains of 1 or less. I don't see what difference it makes
whether the reference signal remains steady forever, or varies
once in a while, or changes continuously.

MartinL:

Inasmuch as "control" is a special case of dynamics, whatever
is true in general of dynamic systems must also be true of
control systems, though the reverse is not the case.

The idea that dynamic systems analysis says everything that
control theory says is a delusion. Nobody has actually come up
with a general model of all dynamical systems. That would be a
model of the entire universe. What we see are specific designs
that are claimed to be general, but which overlook such general
problems as non-zero non-stochastic disturbances and variations
in local properties of the environment -- as well assertions that
a specific design covers all possible designs.

If there were any statements that are true of all possible
dynamic systems, then of course they would apply to control
systems, too. But such general statements also have to apply to
systems that are not control systems, so they would be unable to
express what is different about a control system and, say, a
marble in a bowl or a billiard ball rolling along a flat table.
That is the trouble with general statements: the more different
cases they cover, the less interesting is what they have to say
about any one case. They say more and more about less and less
until they say everything about nothing at all. We don't need a
general theory of all dynamic systems; we need a theory that
deals with the kinds of dynamic systems we find in the form of
organisms. And that theory is control theory.
-----------------
We're arguing about angels dancing on pinheads. Fooey on this.
------------------------------------------------------------
Best to all,

Bill P.

[From Bill Powers (951123.1430 MST)]

Martin Taylor 951123 13:30 --

     Exactly. In the point being illustrated, the "position of the
     switch" IS fixed by Nature. As soon as you permit the
     "experimenter" to move it, you void the analogy and evade the point
     of the example.

Why is there a rule against the experimenter's manipulating the
environment as a way of doing the Test? I can see that imposing such
rules would make it impossible to apply the test to distinguish between
the variable the system is controlling and the actually effective
variable. If we can _see_ that there's a difference, why make a rule
against making the appropriate changes to apply the appropriate
disturbance? And if we _can't_ see the difference, how will we ever know
that this problem exists? What is the point you're trying to make? That
we don't perceive reality directly?

I think we're getting into angels-on-pinheads territory.

···

-----------------------------------------------------------------------
Best,

Bill P.

[Martin Taylor 951123 17:45]

Bill Powers (951123.1430 MST)

Martin Taylor 951123 13:30 --

    Exactly. In the point being illustrated, the "position of the
    switch" IS fixed by Nature. As soon as you permit the
    "experimenter" to move it, you void the analogy and evade the point
    of the example.

Why is there a rule against the experimenter's manipulating the
environment as a way of doing the Test?

Because we are not illustrating an experimenter doing the Test. We are
illustrating the development of the ability of the person to turn on
lights in rooms, and by extension the ways in which people learn to
perceive. Nature puts the switch where it is, and if Nature wants
to do the Test on the jacket-flipping person, Nature is quite at liberty
to move the switch (change the laws of physics, for example). But we don't
usually give such scientific attributes to Nature (though the Bible does
talk about God testing some people in analogous ways).

What is the point you're trying to make? That
we don't perceive reality directly?

Not at all. It is the same point I made from one direction when I discussed
the reorganizations of Jack-the-cat who has been in a variety of different
boxes. The jacket-flipper can turn on the light, but not if he dances in
with his toes turned out. It takes only a _very_ small variation in the
ritual to cause it to fail, and therefore it is highly improbable that the
flipper will ever discover the effective part of the action. He will be
able to turn on the light in this room, by using the correct ritual, which
he can teach to his students and descendants. The ritual becomes a moral
imperative, performed by all "right-thinking" people. It won't apply in
other rooms, so only bad people will go into those rooms. Scientists
investigating them will be excoriated for breaking the taboo.

But if the flipper can find ways of varying the actions and _yet keep the
ability to turn on the light_ it is quite possible that the consequential
action will become part of a loop that directly involves a controlled
perception rather than being a side-effect that happens to complete the
outer "see the light as lit" control loop. That's known as "error-free"
learning ("error" here not being the same as PCT "error," of course). If
there is an easily followed continuum of changes of action that can always
maintain the outer loop's control, it is not at all unlikely that the
jacket-flipper can turn into a calm, easy switcher-on of lights, in this
and in other rooms. The jacket-flip might turn into a brush with the wrist,
the dance steps might become smaller and disappear, the wrist might give
way to fingers...

The ex-jacket-flipper will happily go into other rooms and turn on the
lights in most of them. No taboo, no ritual.

I think we're getting into angels-on-pinheads territory.

These angels fly. The pinheads stay behind. (Hamlet, or something like that).

Martin